Research map / 3D vision / Face and shape
Pixel2Mesh++: Multi-View 3D Mesh Generation via Deformation
Chao Wen, Yinda Zhang, Zhuwen Li, Yanwei Fu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add feature statistics (mean, max, std) instead of concatenation (ablation) | Pixel2Mesh++ MDN · ShapeNet multi-view | F-score(tau): 65.26 -> 66.48 | Improved |
| add re-sampled Chamfer loss instead of vertex-only Chamfer loss (ablation) | Pixel2Mesh++ MDN · ShapeNet multi-view | F-score(tau): 66.26 -> 66.48 | Improved |
| add re-sampled Chamfer loss instead of vertex-only Chamfer loss (ablation) | Pixel2Mesh++ MDN · ShapeNet multi-view | Chamfer distance: 0.496 -> 0.486 | Improved |
| add feature statistics instead of concatenation (ablation) | Pixel2Mesh++ MDN · ShapeNet multi-view | Chamfer distance: 0.511 -> 0.486 | Improved |
| add MDN on top of MVP2M coarse shape | MVP2M · ShapeNet multi-view | F-score(tau): 61.05 -> 66.48 | Improved |
| add MDN on top of P2M-M coarse shape (ablation) | P2M-M · ShapeNet multi-view | F-score(tau): 61.72 -> 64.47 | Improved |
| add MDN on top of P2M-M coarse shape (ablation) | P2M-M · ShapeNet multi-view | Chamfer distance: 0.548 -> 0.493 | Improved |
| use MVP2M instead of P2M-M as initialization for MDN (ablation) | MDN · ShapeNet multi-view | F-score(tau): 64.47 -> 66.48 | Improved |
| increase test views from 2 to 5 (trained with 3 views) (ablation) | MDN · ShapeNet multi-view | F-score(tau): 64.48 -> 68.29 | Improved |
| increase test views from 2 to 5 (trained with 3 views) (ablation) | MDN · ShapeNet multi-view | Chamfer distance: 0.515 -> 0.459 | Improved |
| train with matching view count (5) instead of fixed 3 views (ablation) | MDN · ShapeNet multi-view (5 test views) | F-score(tau): 68.29 -> 68.82 | No clear effect |
| train with 2 views instead of 3 views (ablation) | MDN · ShapeNet multi-view (2 test views) | F-score(tau): 64.48 -> 64.11 | No clear effect |
Closest papers by meaning
- Pixel2Mesh++: 3D Mesh Generation and Refinement From Multi-View Images (TPAMI 2022 · similarity 0.99)
- Pixel2Mesh: 3D Mesh Model Generation via Image Guided Deformation (TPAMI 2020 · similarity 0.94)
- Geometry-Aware Network for Non-Rigid Shape Prediction From a Single View (CVPR 2018 · similarity 0.90)
- Synthesizing 3D Shapes via Modeling Multi-View Depth Maps and Silhouettes With Deep Generative Networks (CVPR 2017 · similarity 0.88)
- SurfNet: Generating 3D Shape Surfaces Using Deep Residual Networks (CVPR 2017 · similarity 0.88)
- Visual Object Networks: Image Generation with Disentangled 3D Representations (NeurIPS 2018 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
